CLASSIFICATION BY DUAL SCALING AND THE REPLACEMENT PROCEDURE

DOI Web Site 16 References

Bibliographic Information

Other Title
  • 双対尺度法と入れ替えのアルゴリズムによる分類
  • ソウツイ シャクドホウ ト イレカエ ノ アルゴリズム ニ ヨル ブンルイ

Search this article

Description

As a descriptive method for multidimensional qualitative data, numerous methods were proposed, for example, Guttman's scaling (Guttman, 1941), quantification method (Hayashi, 1952), correspondence analysis (Benzecri, 1973), dual scaling (Nishisato, 1980), etc. These methods aim to give a scale for qualitative data and to describe data structure. Considering the classification of qualitative data, we partition variables or individuals by two steps, to give a scale for qualitative data by the methods mentioned above, to partition variables or individuals by the scale. However the scales obtained by these methods are designed for describing data, and not designed for classification of data. In this paper, we propose the method for finding the partition of individuals or variables and scale for classification simultaneously. This method combines dual scaling and replacement procedure. The replacement procedure uses the algorithm in the clustering method based on homogeneity (Nakamura & Baba, 1999). Applications of this method for data are also discussed.

Journal

References(16)*help

See more

Keywords

Details 詳細情報について

Report a problem

Back to top